Update run_eval.py

run_eval.py

@@ -163,9 +163,11 @@ for cfg in CONFIGS:
     count_before = len(all_rows)
     for task, scores in res["results"].items():
         for metric, value in scores.items():
-            if metric not in METRICS_TO_KEEP:
+            base_metric = metric.split(",")[0]
+            if base_metric not in METRICS_TO_KEEP:
                 continue
             all_rows.append({**meta, "task": task, "metric": metric, "value": value})
+
     print(f"{len(all_rows) - count_before} rows added for {adapter_repo}")